📚 Quantiles and k-values | 分位数与k值
In statistics, quantiles are cut points dividing a dataset into groups of equal frequency. The term “k-value” often refers to the order number of a particular quantile or the critical value from a standard normal distribution used in confidence intervals and tolerance limits. Understanding quantiles and k-values is essential for summarizing data distributions, detecting outliers, and making inferences about populations. This article covers the core concepts required in the IB Mathematics curriculum, including quartiles, percentiles, the interquartile range, and the use of critical k-values in normal-based estimation.
在统计学中，分位数是将数据集分成等频数分组的切割点。“k值”一词通常指特定分位数的次序编号，或标准正态分布中用于置信区间和容忍界限的临界值。理解分位数与k值对于概括数据分布、检测异常值以及对总体进行推断至关重要。本文涵盖了IB数学课程要求的核心概念，包括四分位数、百分位数、四分位距以及基于正态估计中临界k值的运用。
1. Introduction to Quantiles and k-values | 分位数与k值简介
Imagine you have a sorted list of exam scores. How can you find the score that cuts off the bottom 25% of students? That cut point is a quantile, and the specific 25% mark is called the first quartile, often denoted Q₁. The generic term “k-th quantile” or “k-th percentile” uses k to represent the proportion of data lying below that value. For example, the 90th percentile is the value below which 90% of the data fall. The concept of k-values extends to statistical tables where a critical value k (or z) is used to construct confidence intervals.
假设你有一组已排序的考试成绩。怎样才能找到截断最低25%学生的分数？该切割点就是一个分位数，而特定的25%标记被称为第一四分位数，通常记作 Q₁。通用术语“第k分位数”或“第k百分位数”使用k来表示位于该值之下的数据比例。例如，第90百分位数是指有90%的数据落在其下的数值。k值的概念还延伸到统计用表，其中临界值k（或z）用于构建置信区间。
2. Defining Quantiles: Splitting Data into Equal Parts | 分位数的定义：将数据等分
A quantile is a generic term for a value that partitions a sorted dataset into groups of roughly equal size. If we divide the data into four equal parts, we get quartiles (Q₁, Q₂, Q₃). Five equal parts give quintiles, ten equal parts produce deciles, and one hundred equal parts yield percentiles. The second quartile Q₂ is the median (the 50th percentile). The k in “k-th quantile” indicates the number of intervals: for m-quantiles, there are m − 1 partition values. In IB Mathematics, the most commonly used quantiles are quartiles and percentiles.
分位数是一个通用术语，指将已排序的数据集划分为大小大致相等分组的数值。如果把数据分成四等份，就得到四分位数（Q₁, Q₂, Q₃）；五等份得到五分位数；十等份得到十分位数；一百等份得到百分位数。第二四分位数 Q₂ 就是中位数（第50百分位数）。“第k分位数”中的k表示区间的数量：对于m分位数，有 m − 1 个分割值。在IB数学中，最常用的分位数是四分位数和百分位数。
3. Quartiles and the Five-Number Summary | 四分位数与五数概括法
Quartiles split a dataset into four equal groups. The first quartile (Q₁) is the value below which 25% of the observations lie. The second quartile (Q₂) is the median. The third quartile (Q₃) has 75% of data below it. Together with the minimum and maximum values, these five numbers form the five-number summary. This summary is the foundation of box-and-whisker plots, allowing quick visual comparison of distributions, symmetry, and spread.
四分位数将数据集分成四个等组。第一四分位数 Q₁ 是25%观测值落在其下的值；第二四分位数 Q₂ 是中位数；第三四分位数 Q₃ 下有75%的数据。这些与最小值和最大值一起构成了五数概括法。这个概要是箱线图的基础，可以快速直观地比较分布、对称性和离散程度。
4. Calculating Quartiles: Different Methods and the k Formula | 四分位数的计算：不同方法和k公式
the two nearest data points. For example, with n = 11, the position for Q₁ is (1/4)×(11+1) = 3, so Q₁ is the 3rd value. Some syllabi also teach the inclusive method or the Moore and McCabe method, but the (n+1) approach is widely accepted.
对于小型离散数据集，可以通过先将n个数据排序来求四分位数。IB中常用的位置公式是：L = (k/4) × (n + 1)，其中对于 Q₁, Q₂, Q₃，k = 1, 2, 3。如果L是整数，四分位数就是该位置上的数据值。如果L不是整数，则在最接近的两个数据点之间进行线性插值。例如，当 n = 11 时，Q₁ 的位置是 (1/4)×(11+1) = 3，因此 Q₁ 是第3个数值。一些课程大纲也教授包含法或Moore与McCabe方法，但(n+1)方法被广泛接受。
Position for k-th quartile: L = (k/4)(n + 1)
第k四分位数的位置公式：L = (k/4)(n + 1)
5. Percentiles: The k-th Percentile | 百分位数：第k百分位数
rcentiles divide data into 100 equal parts. The k-th percentile is the value below which k% n integer, interpolation is again used. Percentiles are useful in standardised testing, growth charts, and when you need a finer division than quartiles. For example, the 95th percentile indicates a value higher than 95% of the population.
百分位数将数据分成100等份。第k百分位数是有k%的数据落在其下的值。对于一个容量为n的已排序数据集，第k百分位数（k从1到99）的通用位置公式是：L = (k/100) × (n + 1)。当L不是整数时，同样使用插值法。百分位数在标准化测试、成长图表以及需要比四分位数更精细的划分时非常有用。例如，第95百分位数表示比总体中95%的值都高的一个值。
Position of k-th percentile: L = (k/100)(n + 1)
第k百分位数的位置：L = (k/100)(n + 1)
6. Interquartile Range and Outlier Detection | 四分位距与异常值检测
rquartile range (IQR) is defined as IQR = Q₃ − Q₁. It measures the spread of the middle 50% extreme outlier criteria (often 3 × IQR). The boxplot visualises these fences with whiskers extending to the farthest non-outlier data point.
四分位距（IQR）定义为 IQR = Q₃ − Q₁。它衡量了中间50%数据的离散程度，并且不受异常值影响。IQR 通过以下规则识别潜在异常值：若某值小于 Q₁ − 1.5 × IQR 或大于 Q₃ + 1.5 × IQR，则被视为异常值。这些界限称为下围栏和上围栏。系数1.5是常规使用的，对于极端异常值常乘以3。箱线图通过让须线延伸到最远的非异常值数据点来直观展示这些围栏。
7. Quantiles for Continuous Random Variables | 连续随机变量的分位数
, which may require inverse CDF functions. In IB, this is commonly seen with the normal distribution where quantiles correspond to given left-tail probabilities. The k-value here is the critical value on the horizontal axis cutting off a specified tail area.
对于连续概率分布，第k分位数 qₖ 是使得累积分布函数 F(qₖ) = P(X ≤ qₖ) = k/100（对于百分位数）或对于概率p有 F(qₖ)=p 的值。例如，中位数满足 F(q₀.₅) = 0.5。求分位数通常涉及解 F(x)=p 得出 x，这可能需要逆累积分布函数。在IB中，这种情况常见于正态分布，其中分位数对应给定的左尾概率。这里的k值是横轴上截断指定尾部面积的临界值。
8. The Standard Normal Distribution and Critical k-values | 标准正态分布与临界k值
The standard normal distribution Z ~ ons from the mean that correspond to a desired confidence level. Common values are: for 90% confidence, z* ≈ 1.645; for 95% confidence, z* ≈ 1.96; for 99% ) = confidence level. Mathematically, z* = Φ⁻¹(1 − α/2) where α is the significance level.
标准正态分布 Z ~ N(0, 1) 在推断中起着关键作用。临界k值，通常记作 z* 或 k，是与期望置信水平相对应的、距离均值的标准差个数。常见的值有：90%置信水平时 z* ≈ 1.645；95%时 z* ≈ 1.96；99%时 z* ≈ 2.576。这些k值是通过逆正态函数获得的，满足 P(−z* < Z < z*) = 置信水平。数学上，z* = Φ⁻¹(1 − α/2)，其中 α 是显著性水平。
Standard normal critical values (k-values): z₀.₉₅ = 1.960, z₀.₉₉ = 2.576
标准正态临界值（k值）：z₀.₉₅ = 1.960, z₀.₉₉ = 2.576
9. Confidence Intervals: Using k-values for Estimation | 置信区间：用k值进行估计
with a known standard deviation σ, a confidence interval is built using the sample mean x̄ and the critical k-value. The formula is: CI = x̄ ± k × (σ/√n), where k is the z* value for the desired confidence. For unknown σ, the t-distribution replaces the normal and the k-value becomes a t-critical value with degrees of freedom n−1. These k-values are directly linked to quantiles of the t-distribution. In IB examinations, you may be required to find such k-values from tables or GDC and interpret the interval in context.
当已知总体标准差σ时，估计总体均值需利用样本均值 x̄ 和临界k值构建置信区间。公式为：CI = x̄ ± k × (σ/√n)，其中 k 是所需置信水平的 z* 值。若σ未知，则用t分布代替正态分布，k值变为自由度为 n−1 的t临界值。这些k值与t分布的分位数直接相关。在IB考试中，你可能需要从表格或图形计算器中查找此类k值，并结合背景解释区间。
Confidence interval for mean: x̄ ± k · (σ/√n)
均值的置信区间：x̄ ± k · (σ/√n)
10. Worked Example: Quartiles and IQR from a Dataset | 例题：从数据集中求四分位数和IQR
34. Then IQR = 34 − 18 = 16. The lower fence = 18 − 1.5 × 16 = −6, upper fence = 34 + 1.5 × 16 = 58. Since all data lie within these fences, there are no outliers.
考虑已排序数据：12, 15, 18, 21, 23, 25, 28, 30, 34, 37, 40。共 n = 11 个值。使用位置公式 L = (k/4)(n+1)，对于 Q₁ (k=1)：L = (1/4)(12) = 3，第3个值是18，故 Q₁ = 18。Q₂ (k=2)：L = (2/4)(12) = 6，第6个值是25，故 Q₂ = 25。Q₃ (k=3)：L = (3/4)(12) = 9，第9个值是34，故 Q₃ = 34。IQR = 34 − 18 = 16。下围栏 = 18 − 1.5 × 16 = −6，上围栏 = 34 + 1.5 × 16 = 58。所有数据都在这些围栏之内，因此没有异常值。
11. Worked Example: Normal Distribution k-value and Confidence Interval | 例题：正态分布k值与置信区间
A random sample of 50 light bulbs has a mean lifetime of 1200 hours. The population standard deviation is known to be 80 hours. Construct a 95% confidence interval for the true mean lifetime. For 95% confidence, the k-value (z*) is 1.96. The standard error is σ/√n = 80/√50 ≈ 11.3137. The margin of error is 1.96 × 11.3137 ≈ 22.17. Thus the 95% CI is 1200 ± 22.17, giving (1177.83, 1222.17). We are 95% . The k-value here is derived from the 0.975 quantile of the standard normal distribution.
一个包含50个灯泡的随机样本，其平均寿命为1200小时。已知总体标准差为80小时。构建真实平均寿命的95%置信区间。对于95%置信水平，k值（z*）为1.96。标准误为 σ/√n = 80/√50 ≈ 11.3137。误差幅度为 1.96 × 11.3137 ≈ 22.17。因此，95% CI 为 1200 ± 22.17，即 (1177.83, 1222.17)。我们有95%的信心认为这些灯泡的总体平均寿命介于1177.83与1222.17小时之间。这里的k值来自标准正态分布的0.975分位数。
12. Summary and Key Takeaways | 总结与关键要点
es, applying the 1.5×IQR rule for outliers, and interpreting z* and t* critical values for confidence intervals. Mastery of these concepts will enable you to tackle IB examination questions on data summary, normal distributions, and statistical inference with confidence.
分位数为描述数据集中位置和离散程度提供了一种稳健的方法。k值中的k可以代表分位数的次序（例如，第k百分位数）或参考分布的临界值。关键技巧包括：对四分位数和百分位数使用(n+1)位置公式，应用1.5×IQR规则识别异常值，以及解读用于置信区间的z*和t*临界值。掌握这些概念将使你能够自信地应对IB考试中关于数据概括、正态分布和统计推断的问题。
Published by TutorHao | Mathematics Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图